To say the past year has been challenging for inventory management would be an understatement. The first wave of the Covid-19 pandemic caused rush buying and stock outs on many types of goods and because of shutdowns in other sectors, it caused overstocks in some categories. Beyond that, the level of e-commerce ordering for many companies has rivaled or exceeded the level of online orders seen during the holiday season.
As a result, many companies have inventory levels that are out of whack. And, it’s beyond making one-time adjustments. Planners need to find a way to plan when history is of limited value. When you add in ongoing challenges like trying to ensure inventory accuracy in warehouses and keeping forward-pick areas slotted effectively, it’s a good time to rethink inventory management approaches.
“The first thing many companies have started doing is that instead of looking at history on a monthly basis, they’ve pivoted their inventory planning based more on weekly sell cycles to try to understand where the trends are going, especially for items that are seeing a demand surge with Covid,” says Aman Sapra, director of global supply chain and inventory strategy with consulting firm St. Onge. “This shorter-term focus gets into what is known as demand sensing, so in effect, folks are looking to ‘sense’ demand more so than just use history.”
Surveys back up the notion that inventory management needs more focus. In Peerless Research Group’s recent “2020 Warehouse and DC Operations Survey,” 69% of respondents said improving inventory control was an action taken to adjust operations since pandemic, ranking only behind adjusting warehouse processes as the top area of adjustment. Another 2020 survey of supply chain professionals by Verusen, a company with a Cloud inventory analytics solution, found that 10% felt fully prepared for Covid-19 impacts.
Inventory management spans both planning and execution, making it relevant to DC operations. While inventory planning typically happens at a corporate level, solutions such as inventory optimization applications are used by some companies to decide how much inventory should be at each DC.
Ultimately, getting inventory back under control will likely involve planning that is more adept at assessing the impact of current demand signals, as well as a higher level of inventory accuracy and visibility across supply chain nodes, including warehouses.
DCs that were well established in the use of execution solutions such as warehouse management system (WMS) software, bar code data collection and mobile industrial devices were in a better place when it came to inventory control during the pandemic than organizations that lagged behind on WMS, notes Norm Saenz Jr., a managing director with St. Onge.
“On the execution side, not a lot has changed in terms of what it takes to keep inventories under control within the four walls,” says Saenz. “Having a high level of inventory accuracy and clear visibility into where inventory actually is within your DCs mainly comes from having a good WMS, and using bar codes and scanning equipment and verification steps as part of that. This has been true for years, but the pandemic has put a greater emphasis on it.”
Effective inventory management given the present challenges involves a two-pronged approach: disciplined use of WMS and supporting data capture technologies; and a fresh approach to inventory planning that puts more weight on short-term demand signals like current order patterns or point of sales (POS) data. The right approach may build on information management solutions that have been around for years, but will tighten things up in several ways to bring inventory accuracy to a higher level.
The demand and inventory situation has varied widely for different industry sectors and companies, notes Sapra, from companies that make or distribute personal protective devices or high-demand consumer goods, to companies that were focused on industries that contracted during the pandemic, like suppliers to restaurants. For companies that have seen a demand surge, the trend is toward leveraging the most current sales/demand data.
“A handful of companies have pivoted to focusing on the core products in their mix,” says Sapra. “They’ve been thinking about getting better demand sensing data and looking at how their core products are moving on a weekly basis, and planning out the next few weeks or months, using that latest data.”
By contrast, Sapra explains, companies serving sectors that have been slower during the pandemic have been trying to single out products that have decent demand and looking at ways to build demand for goods that are moving slowly. “Instead of inventory planning based on history, they’re saying: Let’s just make sure we’re staying inbounds on inventory for products that are selling, while trying to incentivize to build demand for the sale of items that have slowed down,” he says.
The common denominator to success is putting more weight on the freshest demand data, rather than trying to plan based on what sold at the same time a year before or in preceding months. Most inventory planning solutions, adds Sapra, have logic and rules engines that would allow a company to put more emphasis on current demand signals, and less on history that predates the pandemic.
Planning logic such as exponential smoothing techniques coupled with machine learning regression techniques are supported by various planning solutions so that essentially, less weight is given to longer-term history.
“There are ways to basically say, ‘I want to bias toward the most recent history in being indicative of the future when planning, rather than being biased toward the more far-out history,” says Sapra.
Another way companies have been looking to improve on inventory management at the supply chain level during the pandemic is to have tighter visibility into supplier shipments using mechanisms like supply chain control towers, adds Sapra. “Visibility windows have gotten stricter,” he says. “It’s important to know exactly when inventory will be available.”
Today’s planning software makes use of artificial intelligence (AI) and machine learning in a way that constantly evolves as the software sees more data, in a way that “tunes” the planning engine so it can identify the trends that are here to stay as well as the outliers, explains Amanda Oelschlegel, pre-sales director at supply chain software provider Blue Yonder.
“While historical data is used for the initial machine learning ‘training,’ real-time data is constantly feeding the engine and corrections are being made,” says Oelschlegel. “Additionally, the longer you feed these engines, the more data aggregates and the ‘outlier demand’ becomes known. This allows for situations like crazy spikes in toilet paper to be reacted upon quickly, because the demand spike is being seen for long enough for the engine to realize it is not a blip on the radar.”
With pandemic impacts now extending into 2021, AI-based planning software has already “learned” how spikes for commodities unfold and can better predict subsequent spikes for commodities. “As we’re seeing now with a recent surge in the pandemic, spikes in commodities are being seen again,” says Oelschlegel. “Because the engine has seen this before, it will be quicker to react this time than before and customers should be able to find toilet paper on their local shelves because everyone and every tool is a bit smarter this time around.”
In terms of supply chain execution, it’s still important to have solutions like control towers to have insight into actual resupply arrivals versus using more standard lead times, Oelschlegel says.
“Visibility and collaboration have become more important than ever,” Oelschlegel says. “Working with all nodes in your supply chain and understanding the constraints that exist at each point will allow companies to focus on how to best optimize at each milestone.”
Cognitive control towers, inventory portals, vendor collaboration within scenario planning are all tools in the market that can bring entire networks together and map out where disruptions are likely to occur. Once an organization can understand where disruptions are likely to occur, educated decision making on how to best leverage the inventory investment that has been made, or will be made, becomes possible, she adds.
Allocation and consumers
The unusual demand patterns seen during the pandemic makes it difficult to leverage much of 2020’s history in planning, agrees Scott Fenwick, senior director of product strategy at Manhattan Associates. As a result, companies in industries such as retail are relying more heavily on current sales plans to drive inventory plans.
“What most experts with our customers have told us is that, for the most part, they aren’t going to rely on history this year,” says Fenwick. “They will certainly look at that history so if other extreme disruptions do occur, there is data on which products were most affected. They may be able to leverage some of that data, even if the volume level doesn’t apply.”
Fenwick says that in sectors such as apparel and footwear, current sales plans can be used as a starting point for allocation, which is the process of deciding how much and what particular inventory to position at DCs and stores. “A sales plan might not be specific down to sizes or colors, but sales plans can serve as a starting point for a new season, and as soon the season starts to take shape and you have data on what is starting to sell, you can use that data to refine the plan.”
Multiple applications can help with inventory management, from inventory optimization software to distributed order management (DOM) software to automate the decision making involved in terms of which DC or store, or combination of locations, is the best location to fulfill orders from.
Manhattan Associates, which offers software for inventory planning and DOM as well as WMS, also recently launched an allocation solution called Manhattan Active Allocation.
The allocation solution is aimed at helping companies in sectors such as apparel and footwear determine how much inventory to place or “allocate” to specific DCs and stores. The solution’s focus is on providing allocators a better understanding of demand by giving them direct insight into omni-fulfillment trends and customer choices—such as buy online, pick up in store (BOPIS), or curbside pickup, or more traditional in store shopping—so the best positioning of inventory takes place given the customer engagement strategy, says Fenwick.
“A good distributed order management solution does a great job of finding the least cost method of fulfilling a customer order, but the allocators goal in today’s world is to figure out ‘how can I better position inventory in the first place,’ based on a granular understanding of the different types of consumer experiences offered, such as curbside pickup,” says Fenwick. “With a detailed understanding of the fulfillment experience consumers want, more often than not, the inventory is already in the right place, or close to the right place, from the start. Companies have to start basing their allocation practices in a way that reflects business strategy and the consumer experiences being provided.”
Generally, with planning or allocation solutions, it’s valuable to have a solution that automates the introduction of new data into the software’s model, so planners don’t have to rework models from scratch, adds Fenwick. “Our solution is automated, in that it is able to take in sales data and react to it immediately,” he says. “The whole principle behind it is to have software that is agile and helps the organization react, rather than needing to have some planner organize additional data and go in and change the model.”
Optimizing inventory across a DC network is one thing, but even when that’s done well, there’s also the question of where to locate goods within each DC, which includes “slotting” items into forward pick locations. With the larger SKU counts typically associated with e-commerce, many DCs could benefit from slotting adjustments to ensure that goods are optimally slotted within each DC for effective picking, says Dan Basmajian, CEO of Optricity, a provider of slotting software.
“You have to make the best possible use of the cubic capacity of your distribution centers,” says Basmajian. “With the rapid growth of e-commerce, in many instances, you’re going to have more products than what you are used to having. If you use dedicated forward pick locations to fulfill orders, you may no longer have enough of those forward pick locations to accommodate each product.”
For DCs looking to free up space in forward pick areas, adjustments might include moving some slower-moving SKUs to a bulk storage location for picking, or reconfiguring individual forward pick locations to be smaller so more SKUs can be packed into the area, even if that means the replenishment cycle is shorter.
Slotting optimization, explains Basmajian, will assess how fast goods are moving to identify when it makes sense to make slotting adjustments, balancing key outcomes desired from slotting such as reducing the travel time for order selectors, having fewer stock outs, and having controlled, efficient replenishment cycles.
Sometimes, tradeoffs have to be made, adds Basmajian, especially when SKU counts and e-commerce picking requirements are growing and space is limited. “For an item with velocity that is slowing down, it may be better to move that item to bulk storage and pick it from there, even if it involves more travel; considering you really don’t need to pick that item very often anymore, it makes sense to provide its prime location to an item that is being picked more frequently,” he says.
Some DCs might use slotting optimization to help reconfigure forward pick locations to be smaller, allowing more SKUs to be located in the same size forward pick area. The tradeoff in doing this, Basmajian says, is more frequent replenishment.
A labor efficiency assessment can determine if the tradeoff is worth it, which might be the case if the labor involved for slightly more frequent replenishment is offset by more fast-moving SKUs in forward pick to make the work of order selectors there more highly efficient.
“Slotting is a multi-dimensional opportunity,” says Basmajian. “One of its goals is to reduce the time it takes for an order selector to get a location, but it’s more than that. It is about achieving goals while taking into account the operation’s constraints—space always being a huge factor. For many, with the current situation, there can be a real turning of the screws on how much space is going to be available for products, so to some extent, there may have to be some tradeoffs.”
Another thing the pandemic has shown, says Basmajian, is that the disconnect that traditionally has existed between managers responsible for DCs and planners and buyers in merchandising who are making company-wide inventory plans and purchase decisions still continues to prove challenging. There needs to be more open and fluid lines of communication between the two to better manage inventory in these rapidly changing times.
Decisions on inventory that are often heavily influenced by discounts for volume buys should also consider constraints and costs at the DC level, such as the cost to store and pick orders, and get e-commerce orders out the door on time. As Basmajian concludes, “The two domains are not as coordinated as they need to be,” he says. “Buyers in merchandising are often not considering the total cost of ownership of inventory, which includes having to warehouse that product. This disconnect has been around for a long time, but it’s a bigger issue now, since many companies are tight on space, and dealing with more product.”